MANUFACTURING ISSUES FOR A HIGH PERFORMANCE CRYSTAL OSCILLATOR

摘要

Manufacturing issues affecting a high performance balanced-bridge controlled crystal oscillator are described. The design process for this oscillator was presented at FCS in 1997. This paper covers design-for-manufacturability decisions relating to that design, as well as equipment and processes that were created specifically for the purpose of manufacturing this oscillator. The oscillator was manufactured in relatively high volumes for a precision oscillator (over 10,000), and many manufacturing techniques that were acceptable for prototyping purposes and low-volume manufacturing became impractical at higher volumes. The process begins with the crystal. Special precision plating techniques are used to hold calibration tolerance to below I part in 10~(6). The crystal parameters are then carefully measured over temperature in an oil bath. The measured crystal data is put through an algorithm to determine necessary manufacturing parameters. A laser-operated bar code engraving system marks the individual crystals. Statistics on large numbers of crystals are recorded in a database and used to adjust the manufacturing process as necessary to maintain extremely tight angle-of-cut tolerances to hold turnover temperatures within a relatively narrow range. The next step is to install the crystal in a "bridge" PC board. This board includes all of the small number of critical oscillator components; basically consisting of the balanced bridge portion of the oscillator circuit. Fixtures made from modified production oscillator parts are used to test and adjust the bridge board, which must be done at the actual oven temperature to be used. The ovens are individually set to the particular turnover point of each crystal. The bridge board test fixture allows the bridge balance adjustments to be made at the operating temperature. Finished bridge boards, with crystals, are then plugged into oscillator boards that are installed in oven assemblies. Full interchangeability is maintained between bridge boards and oscillator boards. Completed "puck" shaped assemblies are installed in an ager system. The aging of each puck assembly is briefly interrupted to make a final turnover temperature determination. This is done by incrementing the set point of the oven controller in the ager in small steps and observing the frequency of the puck assembly under test. The exact turnover temperature is determined and errors such as thermistor tolerance are calibrated out. This information is added to the database that is used later to store the oven set point data in nonvolatile RAM on the controller board that holds the puck assembly. The ager system also does a simulated holdover test that weeds out most holdover failures before additional resources are expended downstream. Tolerancing issues of the extremely high thermal gain oven are described. Great attention must be paid to the fabrication of the flexible circuit heaters for the oven mass. Another critical parameter is the ratio of heat applied to the top and bottom of the oven mass vs. the rim. We elected to choose a single value used for all ovens, which limited the thermal gain achieved in production to the neighborhood of about 100,000. A technique is described that we could have used, if necessary, to individually set the thermal gains of the ovens to approach 1,000,000.